Optical spectrometry may be performed by a variety of techniques, including multiplex filtering, interferometers and dispersive optical devices. Spectrometers may include, for example, grating-based spectrometers, scanning Fourier transform spectrometers, and dispersive Fourier spectrometers. Grating-based spectrometers typically combine a number of bulk optics components including mirrors, lenses, gratings, optical gratings, apertures, and beamsplitters. Fourier transform spectroscopic technologies are often implemented by using a variant on a Michelson interferometer, with one of the mirrors being scanned in distance along an optical path. The resultant interferogram at the detector is the Fourier transform of the optical spectrum, and the optical spectrum may be recovered by performing a Fourier transform of the received time series of optical intensity data.
Amongst the practical applications of these techniques is Raman spectroscopy. When light is scattered from a molecule or crystal most photons are elastically scattered, having the same energy (frequency) and therefore, the same wavelength, as the incident photons. However, a small component (approximately 1 in 107 photons) is inelastically scattered, at wavelengths that are shifted from the incident radiation. The inelastically scattered photons provide chemical and structural information that is uniquely characteristic of the substance being irradiated. High-resolution detection of this Raman-scattered energy normally requires extensive laboratory facilities and large spectrometer systems, which act as either monochromators or interferometers. Such devices are generally not suitable for portable applications, being precision optical instruments and having moving or adjustable components.
Herein, the terms frequency and wavelength are used to describe the spectral characteristics of an energy spectrum, and a person of skill in the art will recognize that they are equivalent representations that are inversely proportional to each other, where the constant of proportionality is the speed of light. The terms will often be used somewhat interchangeably, so as to permit comparison with various conventional representations of bandwidth, resolution, and the like.
Raman spectra are typically expressed in wave numbers, which have units of inverse length. In order to convert between spectral wavelength and wave numbers of shift in the Raman spectrum, the following formula can be used:
            Δ      ⁢                          ⁢      w        =          (                        1                      λ            0                          -                  1                      λ            1                              )        ,where Δw is the Raman shift expressed in wave number, λ0 is the excitation wavelength, and λ1 is the Raman spectrum wavelength. Most commonly, the units chosen for expressing wave number in Raman spectra is inverse centimeters (cm−1). Since wavelength is often expressed in units of nanometers (nm), the formula above can scale for this units conversion explicitly, giving
            Δ      ⁢                          ⁢              w        ⁡                  (                      cm                          -              1                                )                      =                  (                              1                                          λ                0                            ⁡                              (                nm                )                                              -                      1                                          λ                1                            ⁡                              (                nm                )                                                    )            ×              10        7            ⁢                        (          nm          )                          (          cm          )                      ,Typically, Raman spectroscopy is performed in the range 200-4000 cm−1. A typical excitation wavelength may be 785 nm or 514 nm; however, the selection of wavelength may be governed by a number of considerations, including the avoidance of excitation of fluorescence in the sample.
Typical devices that produce interferograms are usually variants of the Michelson interferometer and generally have moving parts that allow small changes to be introduced in the optical path length between beams of light. An energy beam may be divided into two and beams which travel different optical paths which may be subsequently recombined in a common region where interference occurs. Since a single wavelength would result in a detected intensity that varies periodically with the optical path length, these variations are called fringes, A simplified example of a prior art Michelson interferometer is shown in FIG. 1.
An interferometer operates, typically, by splitting energy from a single source into two beams, and causing one of the beams to travel a different physical distance than the other. When the two beams are brought together again, the phase difference between the beams results in an interference pattern comprised of a series of alternating light and dark fringes, depending on the energy wavelength and the difference in path length, resulting in a variation of detected intensity which is also dependent on the overall spectral characteristics of the energy within the passband of the instrument.
In this example, a Michelson interferometer may comprise four “arms”. The first arm is a source 15 of optical energy, the second arm contains a stationary reflector 20, the third arm contains a movable reflector 25, and the fourth arm leads to an optical power detector 30, such as a photodetector. At the intersection of the four arms an optical beamsplitter 35 is disposed so as to transmit half of the energy impinging thereon and to reflect the other half of the energy. As a result, the energy transmitted by the beamsplitter strikes the fixed reflector 20, and the light reflected by the beamsplitter strikes the movable reflector 25. After reflecting off their respective reflectors, the two energy beams recombine at the beamsplitter 35, and then exit along the fourth arm to an energy detector 30. In this configuration 50 percent of the light is lost prior to reaching the detector.
In a Michelson interferometer 10, a varying path difference between the two beams may be introduced by translating the movable reflector 25 towards and away from the beamsplitter 35. This path difference may be expressed as a phase difference, where the phase difference is proportional to the path difference and inversely proportional to the wavelength of the energy. When the beams that have reflected off the fixed 20 and movable 35 reflectors recombine at the beamsplitter 35 are in phase, an intense beam leaves the interferometer as a result of constructive interference. When the fixed and movable reflector beams are recombined at the beamsplitter 35 so that the beams are out of phase, little energy leaves the interferometer 10 as there is destructive interference. The measured beam intensity at the detector 30 represents the contribution of all the energy from all of the wavelengths that are present. When the reflector 25 is moved so as to change the difference in path lengths of the beam components, the variation of the beam intensity with path difference is termed an interferogram.
Considering the interferogram to be related to the time-domain behavior of the signal resulting from the path length change, the interferogram has been recognized as the Fourier-transform pair of the frequency spectrum of the energy producing the temporal pattern.
Modern digital signal processing technology enables rapid and precise determination of the corresponding frequency spectrum, including the amplitudes of the frequency components, from a time series. Such processing is generally performed by an algorithm known as a Fast Fourier Transform (FFT), although other spectral processing algorithms such as a DFT (discrete Fourier Transform) or Multiple Signal Classification may be used as well. The interferogram may be appodized (weighted) so as to minimize the effects of data truncation, as is known in the signal processing art.
The Michelson interferometer 10 uses a beamsplitter 35 and a moving reflector 25. Changes in the alignment of the beamsplitter 35, and non-uniformities in the movement of the reflector 25 contribute to errors in measurement and reduced repeatability.